Acupuncture for Neck Pain

8/16/2019 Acupuncture for Neck Pain

1/11

Efficacy and safety of acupuncture for chronic uncomplicatedneck pain: A randomised controlled study q

Jorge Vas a,*, Emilio Perea-Milla b, Camila Méndez c, Cayetana Sánchez Navarro d,José Marı́a León Rubio e, Mauricio Brioso a, Inmaculada Garcı́a Obrero f

a Pain Treatment Unit, Dos Hermanas ‘‘A’’ Primary Healthcare Centre, Dos Hermanas, Sevilla, Spainb Research Support Unit, Hospital Costa del Sol, Red IRYSS, Marbella, Má laga, Spain

c Andalusian Public Health Service, Sevilla, Spaind Specialised High Resolution Centre, Rehabilitation Service, Hospital de Valme, Dos Hermanas, Sevilla, Spain

e Department of Social Psychology, School of Psychology, Universidad de Sevilla, Sevilla, Spainf

Rehabilitation Service, Hospital Virgen del Rocı́o, Sevilla, Spain

Received 15 September 2005; received in revised form 31 May 2006; accepted 6 July 2006

Abstract

Chronic neck pain is highly prevalent. To determine the efficacy and safety of acupuncture, in comparison with transcutaneousnerve stimulation-placebo (TENS-placebo) in the treatment of chronic uncomplicated neck pain, a single blind prospective studywas designed, to be carried out at a Primary Healthcare Centre, with random assignment to two parallel groups and with evaluationand analysis by independent evaluators. A random assignment was made from 123 patients of the 149 initially recruited. Thesepatients had been diagnosed with uncomplicated neck pain and experienced neck motion-related pain intensity equal to or exceeding30 on a visual analogue scale (VAS) from 0 to 100 mm. The treatment with acupuncture was compared with TENS-placebo, appliedover 5 sessions in three weeks. The primary endpoint was the change in maximum pain intensity related to motion of the neck, oneweek after the final treatment. Sensitivity was analysed per protocol (PP) and variant analyses were by intention to treat (ITT).Adjustment was made for confounders by multiple linear regression, including baseline values and rescue therapy. By ITT analysis,the change in the pain-VAS variable was greater among the experimental group (28.1 (95% CI 21.4–34.7)). The improvements inquality of life (physical aspect), active neck mobility and reduced rescue medication were clinically and statistically significant. Inthe treatment of the intensity of chronic neck pain, acupuncture is more effective than the placebo treatment and presents a safetyprofile making it suitable for routine use in clinical practice. 2006 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

Keywords: Acupuncture; Placebo; TENS; Neck pain; Randomised controlled study

0304-3959/$32.00 2006 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

doi:10.1016/j.pain.2006.07.002

q Conflict of interest statement: There are no conflicts of interest. Authors’ contributions: Conception and design, J. Vas. Analysis and interpretationof the data, E. Perea-Milla, C. Mendez, J. M. León Rubio. Drafting of the article, J. Vas, E. Perea-Milla, C. Mendez. Critical revision of the articlefor important intellectual content, J. Vas, E. Perea-Milla, C. Mendez. Final approval of the article, J. Vas. Statistical expertise, E. Perea-Milla,C. Mendez. Obtaining of funding, J. Vas. Administrative, technical and logistic support, J. Vas. Collection and assembly of data, C. Sanchez, M.Brioso, I. Garcia Obrero. Grant support: The study protocol was developed in 2001 and was funded by the Consejeria de Salud de la Junta deAndalucia (File No. 52/02) in 2002, and partially by the IRYSS network (File No. G03/202). Recruitment began in 2002 and was completed in 2004.

* Corresponding author. Present address: Unidad de Tratamiento del Dolor, Centro de Salud Dos Hermanas A, Segovia s/n. 41700 DosHermanas, Spain. Tel.: +34 954729811.

E-mail address: [email protected] (J. Vas).

www.elsevier.com/locate/pain

Pain 126 (2006) 245–255


2/11

1. Introduction

Neck pain affects 30% of men and 43% of women atsome time in their lives (Borghouts et al., 1998), and isone that becomes chronic among 10% of men and 17%of women. It is the second most common musculoskele-

tal-related cause of invalidity (Makela et al., 1991;Bovim et al., 1994); the prevalence increases with age,and is greater among women than among men (Coteet al., 2004).

Most cases of chronic neck pain are caused bymechanical factors; although they are usually consideredto be related to degenerative cervical processes, no con-nection has been proven between the patient’s complaintand radiological observations (Harrison et al., 2002).Frequently, non-specific neck pain, with no obviousstructural cause, is recorded (Bogduk, 1984).

The conservative treatment for uncomplicated neckpain includes physical and pharmacological measures,

together with the provision of advice and manual treat-ments such as mobilisation and manipulation (Irnichet al., 2001). Nevertheless, various systematic reviewshave shown that the effectiveness of such interventionsis open to question (Aker et al., 1996; Gross et al.,2000; Dziedzic et al., 2005). Acupuncture is frequentlyapplied to treat chronic neck pain, but despite theincrease in randomised clinical trials (RCT) there is littleevidence for the effectiveness of acupuncture as a treat-ment for the relief of chronic neck pain (Morey, 1998;White and Ernst, 1999; Smith et al., 2000; White et al.,2002). Systematic reviews have revealed that this lack

of evidence is often due to methodological deficienciessuch as too small samples, inadequate methods of statis-tical analysis and the possible physiological activity of placebo (sham) acupuncture (Spitzer et al., 1995;Beaudreuil and Gallou, 2004; Vickers, 2004; Weinerand Ernst, 2004; Paterson and Dieppe, 2005). Moreover,this absence of evidence might arise from the inclusionin some of the reviews of tests examining a wide varietyof painful pathologies (White et al., 2002).

In this paper, we present the results of a controlledrandomised study to evaluate the efficacy and safety of acupuncture in comparison with transcutaneous nervestimulation-placebo (TENS-placebo) for the treatmentof chronic uncomplicated neck pain with a durationexceeding three months.

2. Methods

2.1. Design of the study

Single blind prospective study with random assignation totwo parallel groups, with blind evaluation and statistical anal-ysis by independent evaluators. The study was carried out atthe Pain Treatment Unit (PTU) at a Primary AttentionHealthcare Centre of the Andalusian Public Health Service

(Spain) over a period of two years, from June 2002 to June2004. The study was approved by the Research Ethics Com-mittee at the reference hospital.

2.2. Participants

All of the general practitioners at the Primary Healthcare

Centres in the municipality were informed of the study andtheir collaboration was requested in recruiting and referringto the PTU patients who met the criteria for inclusion in thestudy. These were outpatients aged 17 years and over whohad been diagnosed with uncomplicated neck pain of overthree months’ duration (Skouen et al., 2002), symptomaticat the time of examination, with a motion-related neck painintensity equal to or exceeding 30 on a visual analogue scaleof 0–100 mm (pain-VAS), and who had not received anytreatment during the week preceding their incorporation intothe study. The patients gave their informed consent to par-ticipate in the study. The exclusion criteria are listed inTable 1.

2.3. Randomisation

The patients were randomly allocated to two groups: anexperimental group that was treated with acupuncture, and acontrol group given TENS-placebo treatment. The assignationto one or other of the groups was carried out with validated,publicly available software (Silva Ayzaguer, 1993), withoutusing blocking, but sealed, opaque envelopes. Only the doctorwho applied the treatment was aware of the group assignmentof each patient, and this doctor did not participate in any of the subsequent evaluation phases. A person with no connec-tion with the study was requested to open the envelopes andassign the patients to their respective groups.

The research team adopted appropriate measures to main-tain the confidentiality of the participating patients’ data,including de-identification of the data contained in the dat-abases created for the analysis.

Table 1Exclusion criteria

Exclusion criteria

Previous treatment with acupuncturePain intensity less than 30 mm on a visual analogue scale (0-100 mm)Specific diagnosis of neck pain classed as neuropathologic, infectious,

inflammatory, neoplasic, endocrine, metabolic or visceral

Cervical fracture or traumatismPrevious spinal surgeryNon-specific feverSevere psychiatric illnessSevere disorder of overall health stateInfectious feverish diseaseSevere or generalized dermatopathyAny sign of malign tumorIncompatibility with the medication described in the protocolOccupation-related lawsuit arising from neck pain

Pregnancy

Prior recommendation for treatment with antineoplasic drugs,corticosteroids, immunosuppressor drugs or opioids

Inability or unwillingness to follow instructions

246 J. Vas et al. / Pain 126 (2006) 245–255


3/11

2.4. Treatment protocols

The treatment consisted of five sessions, applied over a peri-od of three weeks, with two sessions during each of the firstand second weeks and one in the third. Both groups were pro-vided with analgesic rescue medication once weekly. The twotreatment modes were applied at different times of the day so

that the patients should not have the opportunity to exchangeimpressions in the waiting room.

2.4.1. Acupuncture

A doctor specialising in acupuncture (accredited by the Bei- jing University of Medical Sciences (China) and by the Scien-tific Society of Medical Acupuncture (ACMAS Huangdi,Spain)), with over 15 years’ clinical experience, selected acu-puncture points on the basis of traditional Chinese treatmentmethods found to be effective for cervical pain (Cobos andVas, 2000; Stux and Pomeranz, 2000) and administered all of the acupuncture treatments. The points were selected accord-ing to the pain characteristics and the accompanying symp-

toms (Table 2). In every case, the puncture was bilateral,with sterile, single-use needles, 25 mm · 0.25 mm or40 mm · 0.25 mm, depending on the area to be treated. Inevery case, the puncture was effected by determining the‘‘Deqi’’ (an elicitation of needle sensation to check that thepuncture was performed in the correct site). The needles werekept in place for 30 min and manually stimulated every 10 min.After removing the needles, Vaccaria seeds were applied in theear auricle at the points described below, after sterilising theskin, and were taped there until the following treatment ses-sion. The patients were told to apply pressure to each ear pointfor a series of 10 repeats three times per day.

2.4.2. TENS-placebo

The TENS-placebo technique applied is a physiologicallyinert type of control treatment that has been used previously(Thorsteinsson et al., 1978; Petrie and Hazleman, 1985;Vincent and Lewith, 1995; Lewith and Vincent, 1996; Woodand Lewith, 1998). The technique was applied usingTRANSMED 911 [Enraf-Nonius B.V.] transcutaneous nervestimulation units that had been adjusted beforehand to pre-vent current passing through the electrodes. With the patientin a prone position, the electrodes were placed at the Jianj-ing (GB 21) bilateral acupuncture point while the nervestimulation unit was located in front of the patient for

30 min such that the flashing diode simulating the stimuluswas both visible and audible. Every 10 min, the patient’sstate was checked and the TENS-placebo potentiometerwas adjusted.

Although the sham intervention was distinguishable fromthe experimental technique, the patients had not received acu-puncture previously and were given this prior information: ‘‘A

research project is being carried out at this Healthcare Centreto evaluate and quantify the effects of two non-pharmacologi-cal techniques for the treatment of chronic neck pain. One willbe applied in a real way, but the other will be false. In a ran-dom way, you will be assigned to one or the other group.You will be given medicines that are considered effective fortreating your neck pain in case the technique applied doesnot relieve it’’.

2.4.3. Rescue medication

The patients in both groups were given a bag containing21 pills (50 mg) of diclophenac for each of the 3 weeks thestudy lasted, with the recommendation to take one (50 mg/

8 h) if pain relief was not obtained with the treatmentgiven. The unused medication was recovered and quantified.Patients with risk factors received gastroprotective drugs.The patients suffering objectifiable muscular contractionwere also given a pill containing tetrazepam (50 mg) every24 h.

2.5. Assessments

The assessments were made blind, by a medical specialist inrehabilitation before the intervention (T0) and one week afterthe final treatment (T1, primary end point). The follow upincluded an assessment at 6 months (week 28) after endingthe treatment (T2, secondary end point). At all times, the

assessors discouraged the patients from commenting uponthe type of treatment received.

2.6. Outcome measures

2.6.1. Primary

Change in maximum pain intensity related to motion of theneck (pain-VAS) assessed at T0 and T1. The patients wereasked to rate the maximum pain intensity related to motionof the neck on a visual analogue scale of 0 (no pain) to100 mm (the maximum pain imaginable).

Table 2Acupoints selected

Main type of cervical pain Proximal points Distal points Ear points

Muscular, miofascial Fengchi (GB 20), Jianjing (GB 21) Taichong (LR 3), Hegu (LI 4),Yanglingquan (GB 34)

Shenmen, Neck, Liver,Muscle Relaxation, Occiput, Thalamus

Arthritic Tianzhu (BL 10), Dazhui (GV 14) Houxi (SI 3) , Shenmai (BL 62) Shenmen, Neck, Kidney,Muscle Relaxation, Occiput, Thalamus

Accompanying symptoms Additional points

Pain in the regionlateral to the neck

Xuanzhong (GB 39)

Anxiety extra YintangDizziness with blurred vision Baihui (GV 20), Sanyinjiao (SP 6)

GB, gallbladder; LR, Liver; LI, large intestine; BL, bladder; GV, governor vessel; SI, small intestine; SP, spleen.

J. Vas et al. / Pain 126 (2006) 245–255 247


4/11

2.6.2. Secondary

The Spanish version of the Northwick Park neck painquestionnaire (NPQ) (Leak et al., 1994) was used to assessthe percentage of discapacity (Gonzalez et al., 2001;Pietrobon et al., 2002) at T0 and T1. One compound indexwas calculated from the sum of the active cervical mobility(ACM) scores in the 6 possible directions of the spinal col-

umn, and another was derived for passive cervical mobility(PCM), measured by goniometry (Gajdosik and Bohannon,1987), with the patient in a seated position, at T0 and T1.To provide a measure of health-related quality of life, thepatients completed the Spanish version of the Short Form-36 Health Survey (SF-36) (Alonso et al., 1995) at T0, T1and T2; in cases of difficulty in understanding items in thequestionnaire, the assessor clarified the doubts. Higherscores indicated a better health-related quality of life. Therescue medication (RM) was quantified at T0, T1 and T2,and categorized into 5 values: 0 (no medication); 1 (less thanthe prescribed medication); 2 (consumption of the prescribeddose); 3 (consumption exceeding the prescribed medication);

4 (consumption exceeding the prescribed dose with non-pre-scribed analgesics). The patients were asked at T0 and T2about visits made to the doctor (general practitioner or spe-cialist) during the previous 3 months (interim medical visits,IMV) as a result of neck pain. At T2, the pain-VAS scorewas assessed as a secondary result measure.

2.7. Treatment credibility

The patients’ degree of confidence was evaluated on aTreatment Credibility Scale (TCS) (Bayreuther, 1994), adaptedfrom one originally proposed by Borkovec and Nau (1972) andvalidated for use in acupuncture research by Vincent (1990),with 4 items scored on a 7-point scale (from 0, no credibility,

to 6, maximum credibility). The first 2 items were scored afterthe first treatment session:

• Do you believe this treatment could help relieve your pain?• Does this treatment seem a logical one?

And the other two items were scored at T1:

• Would you recommend this treatment to a friend or relative suffer-ing the same problem?

• Do you believe this treatment might be suitable for dealing withother problems?

2.8. Statistical analyses

The sample size was determined on the basis of a pre-vious pilot study, taking into account the expected differ-ences between the treatment groups, as expressed bytheir mean scores on the pain-VAS scale one week afterthe treatment. The mean expected difference was 26.5(SD 19) for the experimental group and 14 (SD 18) forthe control group. For a power of 90% and a level of sta-tistical significance of 5%, 49 patients were thus needed forthe experimental group and 46 for the control group.Assuming a loss rate of 20%, the minimum sample sizewas estimated to be 112, this calculation being made withthe freely distributed DSTPLAN software package (Brownet al., 2000).

The values of the sociodemographic and baseline severityvariables for the experimental and control groups were com-pared using measures of dispersion (standard deviation) andof central trend (means). For the primary and secondary resultvariables, a sensitivity analysis was carried out by comparingthe results obtained from the per protocol (PP) analysis andfrom that of intentionto treat (ITT), with imputation of missing

data by the worst case method for the experimental group andthe best case method for the control group (WBC: when datawere missing for a patient in the experimental group, the worstvalue observed at the end of the follow up of this group wasimputed to this patient; when data were missing for a patientin thecontrolgroup, thebest value observed at theend of thefol-low up of this group was imputed to this patient).

Bivariate results for the pain-VAS and for the secondaryresult variables are presented by PP analysis and by ITTaccording to the WBC method (Sackett et al., 2000; EMEAThe European agency for the evaluation of medical products2001). The magnitudes of the differences, for the normally dis-tributed continuous data, were compared using the unpaired

Student’s t test. Skewed data were compared using theMann–Whitney test. The chi-squared test was used to analysethe rescue medication variable. The level of statistical signifi-cance was assumed to be p < 0.05.

The effect size for the raw change scores was evaluated asthe difference of mean differences (mean change in controlgroup subtracted from that of the experimental group) dividedby the pooled standard deviation of the respective changes(Cohen, 1988), calculated by means of a freely distributed soft-ware package (Coe, 2005). The level of credibility was analysedby calculating the mean and its SD for the first two questions(after the first treatment session) and the last two questions (atthe end of the follow up), separately, and by comparing thetreatment groups.

To adjust for possible confounders and to detect potentialinteractions, multiple linear regression models were construct-ed for the result variables, adjusted by the baseline level andusing ITT analysis by the WBC method. The group variables(experimental/control) and baseline variables were includedin the above, together with new variables, concerning socio-demography (age and sex) and the severity of the process(IMV, past duration of the pain, radiological results, type of pain, NPQ, RM, ACM, PCM, SF-36), using criteria of statis-tical significance and of confounding (when the introduction of a new variable produced a change in the coefficient of the var-iable ‘‘group’’ exceeding 10%). The models were constructedagain, but this time removing those observations with Cook’sdistances exceeding the 90 percentile of the distribution, inorder to test the consistency of the results. The assumptionsof the linear models were tested by a normal probability plot(normality) and by plotting studentized residuals againstobserved values (constant variance).

3. Results

3.1. Randomisation, progress through the trial and

baseline data

Of the 149 patients selected, 15 were rejectedbecause they did not meet the criteria for inclusion

248 J. Vas et al. / Pain 126 (2006) 245–255


5/11

(80% of these patients did not record a score of 30 mm on the baseline pain-VAS); of the remaining134, eight refused to participate in the study andanother three were unable to attend the treatment ses-sions. Thus, 123 patients were finally randomised forinclusion in the study. Fig. 1 shows the patients’ pro-

gress during the study and the numbers withdrawingfrom it. The dropout rate was similar in the twogroups, with 3/61 (4.9%) in the experimental groupand 5/62 (8.1%) in the control group (v2 test p = 0.50) being lost.

The baseline characteristics of the samples werefound to be balanced among both groups, concerningthe patients’ characteristics before treatment and withrespect to the dropout rates (Table 3).

3.2. Primary outcome measure

The PP analysis between T0 and T1 for the change inthe pain-VAS variable was greater in the experimentalgroup (44.1, SD 19.5) than in the control group (12.3,SD 14.6), with the differences being statistically signifi-

cant in all the subgroups (Table 4).Table 5 shows the results of the ITT analysis by theWBC method. The difference in the means was 28.1, andthe standardised effect size was 1.51 (CI 95% 1.10–1.90).

3.3. Secondary outcome measures

Table 5 shows the results of the ITT analysis of thesecondary outcome measures one week after treatment.

Fig. 1. Flow of participants through each stage and dropouts.

J. Vas et al. / Pain 126 (2006) 245–255 249


6/11

All the changes in the variables were higher and statisti-cally significant among the experimental group, exceptthe change in PCM and in the Mental Component Sum-mary (MCS) of SF-36.

With respect to consumption of analgesic medica-tion during the experimental period, at T1, 70.5% of the patients in the experimental group had taken norescue medication, versus 17.7% of the control group,compared with all the other categories (RR = 4.0, CI95% 2.3–7.0). In the acupuncture group, 8 patients(29.6%) consumed tetrazepam, while in the controlgroup, this was consumed by 19 (67.9%) of the

patients. Of the latter, 5 (26.3% of those who con-sumed it) took more than the prescribed dose. Thiscircumstance was not observed among the experimen-tal group.

At 6 months after ending the treatment (T2), thechanges in the result variables with respect to baselinevalues had decreased in all cases, and only thepain-VAS changes remained statistically significant(Table 6). Of the 17 patients who were lost to the studyduring the follow up of the control group, 82.3% hadtaken the prescribed dose of rescue medication or hadexceeded it, since T1; in contrast, the patients who left

Table 3Characteristics of the patients before treatment

Outcome measurea Acupuncture group(n = 61)

Control group(n = 62)

Adherers(n = 115)

Dropouts(n = 8)

p value(Adherers and Dropouts)

Age (years) 46.0 (13.7) 47.4 (12.8) 47.2 (13.1) 40.3 (14.3) 0.152b

Sex, numbers (percentages) of patientsMen 15 (24.6) 7 (11.3) 20 (17.4) 2 (25.0) 0.587c

Women 46 (75.4) 55 (88.7) 95 (82.6) 6 (75.0)

Pain duration (months) 47.4 (60.3) 43.0 (40.8) 44.3 (51.8) 58.0 (42.4) 0.466IMV 3.6 (2.8) 4.8 (3.5) 4.2 (3.2) 4.1 (3.4) 0.943

Predominant Rx signs, numbers (percentages) of patientsArthritis 34 (55.7) 34 (54.8) 67 (58.3) 1 (12.5) 0.004Rectification 18 (29.5) 20 (32.3) 35 (30.4) 3 (37.5)Non-specific 9 (14.8) 8 (12.9) 13 (11.3) 4 (50.0)

Type of pain, numbers (percentages) of patientsArthritic 34 (55.7) 34 (54.8) 67 (58.3) 1 (12.5) 0.012Muscular 27 (44.3) 28 (45.2) 48 (41.7) 7 (87.5)

Frequency of pain, numbers (percentages) of patientsOccasional 1 (1.6) – 1 (0.9) – 0.142Frequent 8 (13.1) 4 (6.5) 11(9.6) 1 (12.5)

Very frequent 3 (4.9) 12 (19.4) 12 (10.4) 3 (37.5)Continuous 49 (80.3) 46 (74.2) 91 (79.1) 4 (50.0)

Type of work done, numbers (percentages) of patientsSedentary 12 (19.7) 23 (37.1) 34 (29.6) 1 (12.5) 0.519Moderately active 30 (49.2) 26 (41.9) 51 (44.3) 5 (62.5)Physically intensive 19 (31.1) 13 (21.0) 30 (26.1) 2 (25.0)

Pain-VAS (0–100 mm scale) 68.7 (14.3) 72.3 (15.4) 71.1 (14.7) 61.3 (16.4) 0.070NPQ (0–100 scale) 52.7 (14.0) 56.5 (13.2) 54.6 (13.1) 54.6 (21.6) 0.996

RM, numbers (percentages) of patientsNone 3 (4.9) – 3 (2.6) – 0.946Occasional 7 (11.5) 6 (9.7) 12 (10.4) 1 (12.5)Prescribed dose 36 (59.0) 30 (48.4) 62 (53.9) 4 (50.0)Above prescribed dose 10 (16.4) 23 (37.1) 31 (27.0) 2 (25.0)

Other analgesic medication 5 (8.2) 3 (4.8) 7 (6.1) 1 (12.5)

ACM (degrees) 240.1 (59.4) 229.8 (59.2) 234.7 (59.0) 237.5 (66.5) 0.139PCM (degrees) 291.1 (47.3) 288.4 (47.4) 288.5 (47.1) 308.1 (47.9) 0.213SF-36 PCS (0–100 scale) 36.7 (9.7) 37.6 (7.9) 37.5 (8.4) 32.7 (13.9) 0.365SF-36 MCS (0–100 scale) 38.7 (13.0) 34.0 (11.4) 35.9 (12.1) 41.6 (16.2) 0.213TCSa (0–6 scale) 4.6 (0.8) 4.5 (0.8) 4.5 (0.8) 4.6 (0.9) 0.734

Values are means (standard deviations) unless stated otherwise.a IMV, visits during the previous 3 months to GP or specialist because of neck pain; pain-VAS, intensity of neck pain; NPQ, Northwick Park neck

pain questionnaire; RM, consumption of rescue medication (one week before inclusion in study); ACM, active cervical mobility; PCM, passivecervical mobility; SF-36, short form-36; PCS, physical component summary; MCS, mental component summary; TCSa, mean of first 2 questions of the credibility score, after first treatment session.b This value correponds to age.c This value correponds to sex.

250 J. Vas et al. / Pain 126 (2006) 245–255


7/11

the experimental group either consumed no RM or didso only occasionally.

The patients’ confidence in the technique (measuredon the TCS) after the first treatment session was similaramong the experimental and control groups (4.6 vs. 4.5,respectively; p = 0.48). The patients, therefore, wereconfident in the treatment being received. On the con-trary, the per protocol analysis of the second block of

questions revealed a statistically significant difference( p < 0.001) between the group treated with acupuncture

(mean = 5.1) and the control group (mean = 3.6). Thisestimate correlated with the pain-VAS results at oneweek after ending treatment (T1) (Pearson correlationcoefficient r = 0.681, p < 0.001).

The adjustment for unbalanced variables at T0 pro-vided estimators of the T0–T1 change that were slightlydifferent from those arising from bivariate analysis,although the same result variables (pain-VAS, HNP,

ACM and Physical Component Summary SF-36)remained statistically significant, as shown in Table 7.

Table 4Baseline to post-treatment changes in pain-VAS in subgroups: per protocol analysis

Subgroups (acupuncture patients/control patients) Acupuncture group Control group p value

Pain-VAS (whole group) 44.1 (19.5) 12.3 (14.6)


8/11

3.4. Side effects

The adverse reactions to treatment were mild andaffected the experimental and control groups to a similardegree. The acupuncture group reported four treatment-related adverse reactions (swelling of the hand, bruising,pain and ulcer of the ear), while there were two adversereactions among the control group (cephalea and aggra-vation of the symptoms).

4. Discussion

4.1. Interpretation of the results

The comparison of the experimental and controlgroups in this study reveals differences that are statisti-cally significant and clinically relevant in terms of effec-tiveness in reducing neck pain, measured on a visual

analogue scale. The relative change in the mean intensityof neck pain (basal-final/basal) among the acupuncturegroup was 62.2% (SD 28.5), compared with 20.4% (SD22.5) among the control group. These differencesremained evident in the sensitivity analysis, with perprotocol and various forms of ITT analysis. Animprovement was also found in all the secondary vari-ables, except MCS SF-36 and PCM.

At six months after ending the treatment, the trend of the results remained in this line, although the effect sizewas reduced, with the difference being statistically signif-

icant only in the case of the pain-VAS variable, the effectsize of which varied from intense (1.51) to moderate(0.54), according to the benchmarks proposed by Cohen(1988). The duration of the treatment was estimated onthe basis of clinical experience at the PTU. The improve-ment observed often decreased over the following sixmonths, but remained acceptable for the patients whoregularly attended the PTU, enabling the services pro-vided to be optimised. Furthermore, the treatment wassafe and produced few, and only mild, side effects.

4.2. Methodological limitations

The groups were homogeneous, according to thebaseline evaluation, which indicates that the randomisa-tion procedure functioned well, although the method of using sealed opaque envelopes is not foolproof.

The control measure adopted was an inert techniquethat produced no stimulus in the patient, unlike othercontrol procedures that have been used, such as shamacupuncture, and is supported by various studies whichshow that the credibility of this technique is comparableto that of acupuncture (Petrie and Hazleman, 1985). Inthe present case, the credibility study, after the firsttreatment session, produced similar results in both

Table 6Changes in the results variables from baseline to the end of the follow up (6 months after end of treatment)

Variablesa Acupuncturen = 45

Controln = 40

Mean difference(95% CI)

p value Standardised effect size(95% CI)

Pain-VAS (0–100 scale) 41.1 (26.9) 26.8 (25.9) 14.4 (2.9 – 25.8) 0.014 0.54 (0.10 – 0.97)SF-36 PCS (0–100 scale) 9.3 (11.0) 5.3 (8.0) 4.0 (0.1 – 8.1) 0.054 0.41 (0.02 – 0.84)SF-36 MCS (0–100 scale) 8.0 (13.5) 5.2 (14.1) 2.8 (3.1 – 8.8) 0.351 0.20 (0.23 – 0.63)

IMV 2.5 (2.6) 1.1 (4.3) 1.5 (

0.1 – 3.0) 0.063 0.40 (

0.03 – 0.83)

RM, numbers of patients (percentages) 0.035b

None 26 (57.8) 12 (30.0)Occasional 13 (28.9) 15 (37.5)Prescribed dose 5 (11.1) 8 (20.0)Above prescribed dose 1 (2.2) 5 (12.5)Other analgesic medication – –

Values are means (standard deviations) unless otherwise indicated.a Pain-VAS, intensity of neck pain; SF-36, Short Form-36; PCS, physical component summary; MCS, mental component summary; ICM, visits

during previous 3 months to GP or specialist because of neck pain; RM, consumption of rescue medication (assessment at 6 months after endingtreatment).bv2 test.

Table 7Intention to treat analysis (WC method) in results variables adjustedby basal variables by multiple linear regression models

Variablesa Coefficient CI 95% p value

1. Basal-final pain-VAS (ITT) 28.474 21.632 – 35.315


9/11

groups, thus confirming the adequacy of the placebo.On the contrary, after the treatment had ended, signifi-cant differences were observed between the two groups,in favour of the experimental group; this is a logical con-sequence of the better results obtained by the lattergroup. Other studies have reported similar results con-

cerning the improvement in groups given acupuncturetreatment (Vincent, 1990) or, if no clinically relevant dif-ference was found, that credibility remained at similarlevels among the two groups until the end of the treat-ment period (White et al., 2004). In our case, the appli-cation of the techniques to both groups by the sameperson made it easier to standardise the procedure; how-ever, although this professional was trained to preventthe patients’ noticing any difference in the treatment giv-en to one group or the other, this might comprise a lim-itation of the study.

Few patients withdrew during the study, and resultswere unaffected, as shown by the sensitivity analysis

with different types of ITT and PP analyses. However,during the subsequent follow up, the differences betweenthe groups diminished, which could be attributed to adilution of the acupuncture effect. On the other hand,evidence of possible selection bias arose during the fol-low up of the cohort, with differences being observedbetween the patients abandoning the study. While inthe experimental group, most of those who left the studywere taking less rescue medication than that prescribed,in the control group the majority of patients whodropped out were consuming higher than recommendeddoses of medication. This suggests an alternative expla-

nation to the hypothesis that the acupuncture effect hadbeen diluted in the six months following treatment, asthe reduction in the variable-change differences mighthave been caused, totally or partially, by differences inthe characteristics of those abandoning each group. Thisquestion should be examined in future studies with amedium-term follow up procedure, as it is reasonableto assume it would also affect patient follow up in stan-dard clinical practice.

An attempt was made to minimize the possibly incor-rect classification caused by the above-mentioned effectby using reproducible questionnaires previously testedfor use in Spain; moreover, great care was taken at alltimes to ensure that the assessors worked blind. Theeffect of possible unbalanced confounders was dealt withby multivariate analysis, which revealed only slight vari-ations in comparison with the raw analysis.

Our original hypothesis was that there is a differencebetween the pain relief achieved with the acupuncturegroup, with respect to the control group, of at least12.5 points on the pain-VAS scale, and this value was,in fact, exceeded. Two variables were found not to bestatistically significant, namely PCM and SF-36 MCS.In the case of the former, the possibilities of significantvariation between pre- and post-treatment states are

minimal, due to the normal limitations of cervical func-tioning, while ACM is more closely related to the painexperienced by the patient. In the case of MCS, psycho-social variables require more time for significant changesto become apparent; moreover, this scale includes itemsthat are not directly related to the treatment in question,

but depend on external social variables. During the 6-month follow up, the MCS measured for the experimen-tal group was found to be better than that of the controlgroup, but this improvement could have been influencedby selection bias.

4.3. Other studies of the question

The latest systematic reviews published (White andErnst, 1999; Smith et al., 2000; White et al., 2002)showed that the studies made to date continue to presentproblems of quality, and the authors of these reviewscall for new, methodologically acceptable studies to be

made. Since the latest review was published, newresearch efforts have taken place in which the highlight-ed deficiencies have been addressed. In the study byIrnich et al. (2001), an inert placebo was used, in thesame number of sessions as in the present study, and sta-tistically significant differences were found between thegroup treated with traditional Chinese acupunctureand another group treated with massage, while no suchdifferences were found between the former and thegroup treated with the placebo. A subsequent analysisin which an adjustment was made for baseline measure-ments by means of covariance analysis showed pain

relief to be greater among the acupuncture group thanamong those given the placebo, suggesting that acu-puncture is beneficial in cases of chronic neck pain,and that this benefit is not achieved with the placebo(Vickers, 2004). White et al. (2004) performed a test withWestern acupuncture techniques and obtained resultsthat, although statistically significant, did not achievea given level of clinical relevance (established by theauthors as representing a difference equal to or exceed-ing 30% between the active treatment and the placebo),as only 12% difference was achieved between the twogroups. The greater difference between the latter studyand the present contribution results from the combina-tion of ear auricle points, together with a balanced selec-tion of proximal and distal points. Available evidence,together with studies carried out with reference to othersites (Vas et al., 2004), favours the external validity of the use of acupuncture in the treatment of chronic pain.Nevertheless, given the variability of the results pro-duced in different studies, we believe it is necessary toinvestigate further the number and selection of acu-points on the basis of a diagnosis that is as accurate aspossible (according to traditional Chinese medicine).In addition to the above, there are differences betweenour study and those by White and Irnich that may have

J. Vas et al. / Pain 126 (2006) 245–255 253


10/11

influenced the lack of similarity between the results. Thefirst of these is the baseline level of pain intensity expe-rienced by the patients. In our case, this was 68.7 mm(SD 14.3) among the experimental group and 72.3 mm(SD 15.4) among the control group, with a differenceof 14 mm in the best of cases with the values derived

from the other two studies. The mean age of the patientsin our study was also lower, as was the duration of theillness. Another circumstance that could have led ourresults to differ from those of the above-mentioned stud-ies is the nature of the organisation where the treatmentwas implemented, as medical care in the AndalusianPublic Health System is totally free of charge and thepatient incurs no expense with respect to the treatmentreceived.

5. Conclusions

In the treatment of the intensity of chronic neck pain,

acupuncture is more effective than the placebo treatmentand has a safety pattern that makes it suitable for stan-dard clinical practice.

Acknowledgements

We thank V. Ramos Dominguez for recommenda-tions concerning the project design and C. MarquezZurita for invaluable assistance during the patient-recruiting stage of the study. We also thank the patientswhose collaboration made this project possible.

References

Aker PD, Gross AR, Goldsmith CH, Peloso P. Conservativemanagement of mechanical neck pain: systematic overview andmeta-analysis. BMJ 1996;313:1291–6.

Alonso J, Prieto L, Anto JM. The Spanish version of the SF-36 HealthSurvey (the SF-36 health questionnaire): an instrument formeasuring clinical results. Med Clin (Barc) 1995;104:771–6.

Bayreuther J. A double-blind cross-over study to evaluate theeffectiveness of acupressure at pericardium 6 (P6) in the treatmentof early morning sickness (EMS). Complement Ther Med1994;2:70–6.

Beaudreuil J, Gallou JJ. Acupuncture et cervicalgies chroniques:analyse critique de la littérature. Revue du Rhumatisme2004;71:721–3.

Bogduk N. Neck pain. Aust Fam Physician 1984;13:26–30.Borghouts JA, Koes BW, Bouter LM. The clinical course and

prognostic factors of non-specific neck pain: a systematic review.Pain 1998;77:1–13.

Borkovec TD, Nau SD. Credibility of analogue therapy rationales. JBeh Ther Exp Pschiat 1972;3:257–60.

Bovim G, Schrader H, Sand T. Neck pain in the general population.Spine 1994;19:1307–9.

Brown BW, Brauner C, Chan A, Gutierrez D, Herson J, Lovato J,Polsley J, Russell K, Venier J. DSTPLAN Calculations for SampleSizes and Related Problems. (4.2). 2000. Houston, The Universityof Texas.

Cobos R, Vas J. Manual de Acupuntura y Moxibustión, vol.1. Beijing: Morning Glory; 2000.

Coe R. Effect size calculator. http://www.cemcentre.org/ebeuk/research/effectsize/Calculator.htm. 1-2-2005.

Cohen J. Statistical power analysis for the behavioral sciences. Hills-dale (NJ): Lawrence Erlbaum Assoc.; 1988.

Cote P, Cassidy JD, Carroll LJ, Kristman V. The annual incidence andcourse of neck pain in the general population: a population-basedcohort study. Pain 2004;112:267–73.

Dziedzic K, Hill J, Lewis M, Sim J, Daniels J, Hay EM. Effectivenessof manual therapy or pulsed shortwave diathermy in addition toadvice and exercise for neck disorders: a pragmatic randomizedcontrolled trial in physical therapy clinics. Arthritis Rheum2005;53:214–22.

EMEA The European agency for the evaluation of medical products.Evaluation of medicines for human use. Points to consider onmissing data. http://www.emea.eu.int/pdfs/human/ewp/177699EN.pdf. 2001. 1-3-2005.

Gajdosik RL, Bohannon RW. Clinical measurement of range of motion. Review of goniometry emphasizing reliability and validity.Phys Ther 1987;67:1867–72.

Gonzalez T, Balsa A, Sainz de MJ, Zamorano E, Gonzalez I, Martin-Mola E. Spanish version of the Northwick Park Neck PainQuestionnaire: reliability and validity. Clin Exp Rheumatol2001;19:41–6.

Gross AR, Aker PD, Goldsmith CH, Peloso P. Physical medicinemodalities for mechanical neck disorders. Cochrane Database SystRev 2000;CD000961.

Harrison DE, Bula JM, Gore DR. Roentgenographic findings in thecervical spine in asymptomatic persons: a 10-year follow-up. Spine2002;27:1249–50.

Irnich D, Behrens N, Molzen H, Konig A, Gleditsch J, Krauss M,et al. Randomised trial of acupuncture compared with conven-tional massage and ‘‘sham’’ laser acupuncture for treatment of chronic neck pain. BMJ 2001;322:1574–8.

Leak AM, Cooper J, Dyer S, Williams KA, Turner-Stokes L, FrankAO. The Northwick Park Neck Pain Questionnaire, devised tomeasure neck pain and disability. Br J Rheumatol1994;33:469–74.

Lewith GT, Vincent C. On the evaluation of the clinical effects of acupuncture: a problem reassessed and a framework for futureresearch. J Altern Complement Med 1996;2:79–90.

Makela M, Heliovaara M, Sievers K, Impivaara O, Knekt P, AromaaA. Prevalence, determinants, and consequences of chronic neckpain in Finland. Am J Epidemiol 1991;134:1356–67.

Morey SS. NIH issues consensus statement on acupuncture. Am FamPhysician 1998;57:2545–6.

Paterson C, Dieppe P. Characteristic and incidental (placebo) effectsin complex interventions such as acupuncture. BMJ2005;330:1202–5.

Petrie J, Hazleman B. Credibility of placebo transcutaneous nervestimulation and acupuncture. Clin Exp Rheumatol

1985;3:151–3.Pietrobon R, Coeytaux RR, Carey TS, Richardson WJ, DeVellis RF.Standard scales for measurement of functional outcome forcervical pain or dysfunction: a systematic review. Spine2002;27:515–22.

Sackett DL, Straus SE, Richardson WS, Rosenberg W, Haynes RB.Evidence-based medicine: how to practice and teach EBM. Lon-don: Churchill-Livingstone; 2000.

Silva Ayzaguer LC. Muestreo para la investigación en ciencias de lasalud. Madrid: Ediciones Dı́az de Santos; 1993.

Skouen JS, Grasdal AL, Haldorsen EM, Ursin H. Relative cost-effectiveness of extensive and light multidisciplinary treatmentprograms versus treatment as usual for patients with chronic lowback pain on long-term sick leave: randomized controlled study.Spine 2002;27:901–9.

254 J. Vas et al. / Pain 126 (2006) 245–255


11/11

Smith LA,OldmanAD, McQuay HJ,Moore RA.Teasing apart qualityand validity in systematic reviews: an example from acupuncturetrials in chronic neck and back pain. Pain 2000;86:119–32.

Spitzer WO, Skovron ML, Salmi LR, Cassidy JD, Duranceau J, SuissaS, et al. Scientific monograph of the Quebec Task Force onWhiplash-Associated Disorders: redefining ‘‘whiplash’’ and itsmanagement. Spine 1995;20:1S–73S.

Stux G, Pomeranz B. Fundamentos de acupuntura. Barcelona: Mas-son; 2000.

Thorsteinsson G, Stonnington HH, Stillwell GK, Elveback LR. Theplacebo effect of transcutaneous electrical stimulation. Pain1978;5:31–41.

Vas J, Mendez C, Perea-Milla E, Vega E, Panadero MD, Leon JM,et al. Acupuncture as a complementary therapy to the pharmaco-logical treatment of osteoarthritis of the knee: randomisedcontrolled trial. BMJ 2004;329:1216–9.

Vickers AJ. Statistical reanalysis of four recent randomized trials of acupuncture for pain using analysis of covariance. Clin J Pain2004;20:319–23.

Vincent C. Credibility assessment in trials of acupuncture. Comple-ment Med Res 1990;4:8–11.

Vincent C, Lewith G. Placebo controls for acupuncture studies. J RSoc Med 1995;88:199–202.

Weiner DK, Ernst E. Complementary and alternative approaches tothe treatment of persistent musculoskeletal pain. Clin J Pain2004;20:244–55.

White AR, Ernst E. A systematic review of randomized controlledtrials of acupuncture for neck pain. Rheumatology (Oxford)1999;38:143–7.

White P, Lewith G, Berman B, Birch S. Reviews of acupuncture forchronic neck pain: pitfalls in conducting systematic reviews.Rheumatology (Oxford) 2002;41:1224–31.

White P, Lewith G, Prescott P, Conway J. Acupunctureversus placebo for the treatment of chronic mechanicalneck pain: a randomized, controlled trial. Ann Intern Med2004;141:911–9.

Wood R, Lewith G. The credibility of placebo controls in acupuncturestudies. Complement Ther Med 1998;6:79–82.

J. Vas et al. / Pain 126 (2006) 245–255 255

Documents

Acupuncture for Neck Pain